Distinct Roles for CXCR6⁺ and CXCR6⁻ CD4⁺ T Cells in the Pathogenesis of Chronic Colitis

Colonic Biopsy Specimens

Biopsy specimens were obtained by endoscopy from inflamed areas of the colon of 6 patients with CD and 10 with UC, with the patients’ informed consent. Samples of normal controls (NC) were taken from 5 patients with colonic polyps and were free of inflammation histopathologically. The mean ± SEM (range) age of the patients with CD was 27.3 ± 3.7 (29 – 42) years, while that of UC was 36.6 ± 3.7 (24 – 61) years and that of NC was 55.8 ± 7.0 (33 – 73) years. Clinical activity was evaluated by serum concentration of C-reactive protein, CD Activity Index (CDAI) for patients with CD, and Lichtiger index (UCAI) for UC. Endoscopic activity was evaluated by Simple Endoscopic Score for CD (SES-CD), and Matts classification score for UC patients, respectively. The disease activity of the patients with CD was mild, as the mean ± SEM (range) of CRP was 1.57 ± 0.68 (0.4 – 4.8) mg/L, CDAI was 144.0 ± 45.6 (24.3 – 350.3), and SES-CD was 22.2 ± 6.2 (6 – 48). The activity of the UD patients ranged from remission to severe, as the mean ± SEM (range) of CRP was 2.12 ± 0.85 (0.1 – 6.7) mg/mL, UCAI was 7.9 ± 1.5 (2 – 15), and Matts score was 2.9 ± 0.2 (2 – 4). Two of the patients with CD were receiving no treatment, and 4 were receiving 5-aminosalicylic acid (5-ASA). Two of the patients with UC were receiving no treatment, and 2 were receiving prednisolone, 6 were receiving oral 5-ASA or sulfasalazine with or without 5-ASA enema. The experimental protocol was reviewed and approved in advance by the ethics committees of Chiba University (Permit number: 697) and the RIKEN Yokohama Institute (Permit number: H17-12).

Animals

BALB/cA and Rag1 ^−/− mice were obtained from CLEA Japan (Tokyo, Japan). The mice were maintained under specific pathogen-free conditions in RIKEN animal facilities until use in experiments at 8 to 12 weeks old. All animal experiments were approved by the Animal Research Committee of RIKEN Yokohama Research Institute (Permit number: 24-005).

T cell Preparation

Colonic lamina propria (LP) lymphocytes were prepared as described previously. Briefly, colonic tissues were treated with Hanks’ Balanced Salt Solutions (Wako Pure Chemical Industries) containing 1 mM dithiothreitol and 5 mM EDTA at 37°C for 20 minutes to remove epithelial cells. The tissues were then minced and dissociated with collagenase solution containing 0.5 mg/mL collagenase (Wako Pure Chemical Industries), 1 U/mL dispase (BD Biosciences), 0.5 mg/mL DNase I (Roche Diagnostics), 2% FCS, 100 U/mL penicillin, 100 µg/mL streptomycin, and 12.5 mM HEPES (pH 7.2) in RPMI 1640 medium (Sigma-Aldrich) at 37°C for 30 minutes to obtain single-cell suspensions. After filtering, the single-cell suspensions were washed with 2% FCS/RPMI 1640, and subjected to Percoll gradient separation. Spleen and mesenteric lymph nodes (MLNs) were mechanically disrupted into single-cell suspensions.

Induction of Colitis by Adoptive Transfer of CD4⁺CD45RB^high T cells and by Retransfer of Colitogenic CD4⁺ T cells

Colitis was induced in Rag1 ^−/− mice by adoptive transfer of CD4⁺CD45RB^high T cells as described previously [11]. Briefly, CD4⁺ T cells were enriched from splenocytes from BALB/c mice by the MACS system (Miltenyi Biotec) with biotin-conjugated anti-CD4 monoclonal antibody (RM4-5; BD Biosciences) and anti-biotin microbeads (Miltenyi Biotec). Enriched CD4⁺ T cells were labeled with FITC-conjugated anti-mouse CD3ε (145-2C11), APC-conjugated anti-mouse CD25 (PC61), and PE-conjugated anti-mouse CD45RB (16A) (all from BD Biosciences), and CD3ε⁺CD4⁺CD25⁻CD45RB^high cells were isolated by cell sorting using FACSAria II flow cytometer (BD Biosciences). The Rag1 ^−/− recipients were each given 1×10⁵ CD4⁺CD45RB^high T cells via the tail vein and were sacrificed at 8 weeks after transfer. In retransfer experiments, CD3ε⁺CD4⁺CD25⁻CXCR6⁻ and CD3ε⁺CD4⁺CD25⁻CXCR6⁺ cells were isolated from colonic lamina propria of Rag1 ^−/− recipients at 8 weeks after adoptive transfer of CD4⁺CD45RB^high T cells by cell sorting using a FACSAria II flow cytometer. CXCL16-Fc fusion protein was used to detect CXCR6-expressing cells. These two populations were retransferred into untreated Rag1 ^−/− recipients.

Flow Cytometric Analysis

Lymphocytes were incubated with a mouse CXCL16-human IgG Fcγ fusion protein or control human IgG Fcγ, and specific binding was detected with biotinylated anti-human IgG Fcγ (eBioscience) in combination with streptavidin-APC-Cy7 (BD Biosciences). To characterize cell populations, lymphocytes were incubated with FcγR (CD16/CD32)-blocking mAb (93; eBioscience), and further stained with the following mAbs: FITC-conjugated anti-mouse CD62L (MEL14; BD Biosciences); V500-conjugated anti-mouse CD3ε (500A2; BD Biosciences); Pacific Blue-conjugated anti-mouse CD4 (RM4-5; BD Biosciences); PE-conjugated anti-mouse CD27 (LG.3A10; BD Biosciences); PE-Cy7-conjugated anti-mouse CD44 (IM7; eBioscience); FITC-conjugated anti-mouse CD43 (S7; BD Biosciences) and Alexa Fluor 700-conjugated anti-mouse CD127 (A7R34; eBioscience). Flow cytometric analysis was performed for the stained cells using a FACSAria II flow cytometer with DIVA software (BD Biosciences). To analyze intracellular cytokiness, cells were fixed and permeabilized using Cytofix/Cytoperm solution (BD Biosciences) and stained with FITC-conjugated anti-mouse IFN-γ (XMG1.2: eBioscience), PE-conjugated anti-mouse IL-17A (TC11-18H10; BD Biosciences). For transcription factors, Foxp3/transcription factor staining buffer set (eBioscience), respectively. The cells were then and T-bet (4B10; eBioscience), and Alexa Fluor 647-labeled RORγt (Q31-378; BD Biosciences).

Q-PCR

Total RNA was isolated using an RNeasy Mini Kit (Qiagen), and aliquots of 1 µg of extracted RNA were subjected to reverse transcription (RT) reaction using ReverTra Ace-α (TOYOBO). Real-time PCR analysis was performed to quantify the Cxcl16 and Cxcr6 mRNA expression levels using the SYBR Green PCR assay on a Thermal Cycler Dice Realtime System (TAKARA BIO). The expression of the target gene determined by RT-PCR was presented as a ratio, normalized to an endogenous reference (Gapdh). The specific primers were: 5′-GGC TTT GGA CCC TTG TCT CTT G-3′ (forward) and 5′-TTG CGC TCA AAG CAG TCC ACT-3′ (reverse) for mouse Cxcl16; 5′-AGA ATT TCT TCC GAC TCC CCG -3′ (forward) and 5′-CAG CTC ATC AAT TCC TGA ACC C-3′ (reverse) for human CXCL16; and 5′-GGA CAT TGG TTG CCT CCC TTA-3′ (forward) and 5′-AAA CAA AGC CTG CCT CAC CAC-3′ (reverse) for human CXCR6.

Immunohistochemistry

For immunohistochemical analysis of human CXCL16 and CXCR6, the biopsy samples were fixed in 1% zinc sulfate/4% formalin (Richard-Allan Scientific). Sections of human mucosa 5-µm thick were deparaffinized, rehydrated, and treated with 0.3% H₂O₂ in PBS for 20 min at room temperature to block endogenous peroxidase activity. The sections were incubated with 5% bovine serum albumin in PBS for 30 min at room temperature and then with goat anti-human CXCL16 polyclonal Ab (R&D Systems), mouse anti-human CXCR6 monoclonal Ab (R&D Systems), or an identical concentration of control goat or mouse IgG, overnight at 4°C. The binding of primary Ab was detected with biotinylated donkey anti-goat or mouse IgG (DAKO) followed by streptavidin-horseradish peroxidase (ABC Elite; Vector Laboratories), visualized with 3,3′-diaminobenzidine (DAKO), and counterstained with hematoxylin (DAKO). Immunostaining of mouse CXCL16 was described previously [12].

Microarray Data Collection and Analysis

Total RNA were prepared from CXCR6⁺CD4⁺ T cells and CXCR6⁻CD4⁺ T cells in the colon of colitic mice, and CD45RB^highCD4⁺ naïve T cells in the spleen of BALB/c mice using a RNeasy Plus Mini kit (Qiagen). RNA was amplified and hybridized on the GeneChip Mouse Genome 430 2.0 Array (Affymetrix), according to the manufacture’s procedure. Expression values were determined with Gene Spring version 11.5 (Tomy Digital Biology). The data has also been submitted to GEO database (accession# GSE45881).

Statistical Analysis

Differences between two groups were analyzed by the Student’s t test, unless otherwise specifically noted. When variances were unequal, the data were analyzed by Mann-Whitney U test. In all analyses, P < 0.05 was taken to indicate significance.

Expression of CXCL16 and CXCR6 is Upregulated in the Inflamed Colon of CD Patients and a Mouse Model of CD Colitis

To gain insight into the pathological relevance of the CXCL16-CXCR6 system, we first investigated the expression of these molecules in inflamed colonic mucosa of patients with IBD. Quantitative PCR (Q-PCR) analysis showed that the expression level of the genes encoding CXCL16 and CXCR6 was significantly increased in the mucosa of CD patients compared to healthy subjects and UC patients (Fig. 1A, 1B). The upregulation of CXCL16 was also confirmed at the protein level by Western blot analysis (data not shown). Furthermore, there was a significant correlation between CXCL16 and CXCR6 expression in CD patients (Fig. 1C). By contrast, there were no statistically significant differences in the expression of these genes between UC patients and healthy subjects (Fig. 1A, 1B). Immunohistochemical studies confirmed that CXCL16 was highly expressed by a fraction of LP cells, most likely myeloid cells such as dendritic cells and/or macrophages because of their polymorphic cell shape with relatively large cytoplasm, in the inflamed mucosa of CD patients (Fig. 1D). In addition, colonic epithelium exhibited moderate CXCL16 expression. On the other hand, CXCR6 expression was observed on small round cells with the appearance of infiltrating lymphocytes in the colon of CD patients but not healthy controls (Fig. 1E).

We further examined CXCL16 expression in a well-characterized mouse model of CD colitis induced by adoptive transfer with CD45RB^high naïve CD4⁺ T cells into immunodeficient Rag1^−/− mice. Similar to the situation in CD patients, Cxcl16 mRNA levels were upregulated in colonic epithelial cells (CEC) and whole colon tissue from the colitic mice compared to healthy Rag1^−/− mice (Fig. 2A). Immunohistochemical staining also demonstrated that CXCL16 protein was expressed by CECs, endothelial cells, and a subpopulation of immune cells in the inflamed colon tissue (Fig. 2B).

We subsequently analyzed distribution of CXCR6-expressing cells using a mouse CXCL16-human IgG-Fc fusion protein [13]. Although splenic CD4⁺CD45RB^high naïve T cells before the adoptive transfer lacked CXCR6 expression, 10.9% of CD4⁺CD45RB^low T cells expressed CXCR6 at intermediate levels (Fig. S1). Those CD4⁺CD45RB^lowCXCR6^int T cells had a CD44⁺CD62L⁻ effector memory phenotype. On the other hand, CD4⁺CD45RB^lowCXCR6⁻ T cells were a heterogeneous population that included cells of both CD44⁻CD62L⁺ naïve and CD44⁺CD62L⁻ effector memory phenotypes (Fig. S1C, S1D).

Most CD4⁺ T cells in colonic LP strongly expressed CXCR6 in colitis model mice at 8 weeks after adoptive transfer (Fig. 2C, 2D). Splenic and MLN CD4⁺ T cells expressed CXCR6 to a lesser extent, and only a small portion of BM CD4⁺ T were are positive for CXCR6 (Fig. 2C, 2D).

CXCR6⁻ and CXCR⁺CD4⁺ T cells Subsets in the Inflamed Colon both Contain Effector and Effector-memory Cells

To further characterize CXCR6-expressing CD4⁺ T cells in the inflamed colon, we analyzed their expression of the activation and the memory markers: CD27, CD43, CD44, CD62L, and CD127. Colonic LP CXCR6⁻ and CXCR6⁺ subsets equally contained effector cells of the phenotype CD27⁻CD44⁺CD43⁺CD62L⁻CD127⁻ (Fig. 3A–C). Furthermore, there was no difference in the proportion of CD44⁺CD127⁺ effector-memory cells between the two groups (Fig. 3D). CD44⁺CD127⁺ effector-memory CD4⁺ T cell population can be subdivided into late and early effector-memory populations based on CD27 and CD62 expression, CD27⁺CD62L⁻ and CD27⁻CD62L⁻ cells representing early and late effector-memory populations, respectively [14], [15]. Colonic LP CXCR6⁻ and CXCR6⁺ subsets predominantly contain late rather than early effector memory cells (Fig. 3D–F). These data together demonstrate that the two subsets are composed of an almost identical proportion of effector and late effector-memory cell types (Fig. 3G). However, this was not the case in the spleen, where the SP CXCR6⁻ subset contained fewer effector and late effector memory cells but more early effector memory cells as well as central memory cells (CD44⁺CD127⁺CD27⁺CD62L⁺CD43⁺) compared with the CXCR6⁺ subset. MLN cells also showed a similar phenotype (Fig. S2). Asequential pathway of CD4⁺ T-cell differentiation, from central memory to early effector-memory and then late effector-memory cells has been proposed based on surface marker expression and the length of telomeres [15]–[17]. During this process, a high antigen load would enhance generation of late effector-memory cells. These observations suggest that the CXCR6-expressing CD4⁺ T cells may be more antigen-experienced and highly differentiated than the CXCR6⁻ subset in SP and MLN.

CXCR6⁺CD4⁺ T cells are Responsible for the Production of Inflammatory Cytokines

In order to determine whether CXCR6 expression on CD4⁺ T cells relates to an effector function, we performed intracellular cytokine staining on SP, MLN and colonic LP cells from colitis model mice at 8 week post-transfer. In the colonic LP, the proportions of IFN-γ⁺ cells and IL-2⁺IFN-γ⁺ cells were higher in the CXCR6⁺ subset than in the CXCR6⁻ subset (Fig. 4A). Considering that the majority of colonic CD4⁺ T cells expressed CXCR6 (Fig. 2C), the source of IL-2 and IFN-γ in the inflamed colon was mainly the CXCR6⁺ subset (Fig. 4B). An in vitro Th1 differentiation assay with CFSE-labeled CD4⁺ naïve T cells showed that CFSE⁻CXCR6⁺ cells preferentially produced IFN-γ compared with CFSE⁺CXCR6⁺ cells and CXCR6⁻ cells (Fig. S3). Additionally, IL-17A-producing cells (IL-17A⁺ and IL-17A⁺TNF-α⁺) were predominantly a CXCR6⁺ subset (Fig. 4C, 4D). Immunohistochemical studies also confirmed expression of IL-17A by CXCR6⁺ cells in colonic mucosa of CD patients (Fig. S4). Thus, even though surface markers other than CXCR6 were quite similar between the CXCR6⁺ and CXCR6⁻ subsets, the cytokine production profile clearly distinguished the two populations. Together, the highly expanded CXCR6⁺ subset seems to mediate chronic inflammatory responses in the effector site by producing abundant Th1 and Th17 effector cytokines.

In SP and MLN, the CXCR6⁺ CD4⁺ T subset contained a greater proportion of IL-2⁺IFN-γ⁺ double-producing cells (Fig. 4A, 4B). Furthermore, the proportion and number of IL-17A⁺TNF-α⁺ double-producing cells was much higher in the CXCR6⁺ subset than in the CXCR6⁻ subset (Fig. 4C, 4D). Of note, more than half of the IL-17A-producing CXCR6⁺ cells co-expressed IFN-γ (Fig. 4E). Intracellular staining of T-bet and RORγt also supported the observation that CXCR6⁺ subset is composed mainly of Th1 (T-bet⁺) and Th17 (RORγ⁺) cells (Fig. 4F). Consistent with the co-expression of IL-17A and IFN-γ, the half of RORγt⁺CXCR6⁺ subset co-expressed T-bet.

CXCR6 Expression on CD4⁺ T cells is not Required for the Accumulation of Colitogenic T cells

Given the upregulation of CXCL16 in the inflamed colon as well as the active role of CD4⁺CXCR6⁺ T cells in cytokine production, the CXCL16-CXCR6 system may be important for the development and persistence of the colonic inflammation. This possibility was directly assessed by the adoptive transfer of CD45RB^high naïve CD4⁺ cells isolated from CXCR6-deficient mice (Cxcr6^Egfp/Egfp mice) into Rag1^−/− recipients. CXCR6-EGFP heterozygous mice (Cxcr6^+/Egfp) were used as positive control donor. The adoptive transfer of the CXCR6-deficient T cells induced a wasting disease associated with increased colon weight to a similar extent the control cells (Fig. 5A, 5B). Histopathological analysis of the distal colon confirmed the development of chronic inflammation in both groups (Fig. 5C). The absence of CXCR6 expression by LP CD4⁺ T cells isolated from recipients of Cxcr6^Egfp/Egfp CD4⁺ cells was confirmed by flow cytometry (Fig. 5D). This observation raises the possibility that colitogenic CD4⁺ T cells may redundantly express other chemokine receptors to migrate into colonic lamina propria. In gene expression profiling, colitogenic CD4⁺ T cells displayed upregulation of 8 chemokine receptors including CXCR6 (Table S1).

We also examined whether CXCL16-CXCR6 axis is involved in the signal transduction for proinflammatory cytokine production. The expression levels of effector cytokine transcripts in the inflamed colon tissues were comparable regardless of CXCR6 expression on the adoptively transferred CD4⁺ T cells (Fig. 5E). Consistently, the production of IL-2 and IFN-γ by CXCR6⁺CD4⁺ T cells were comparable when the cells were cultured in Th1-conditioned medium with and without supplementation of recombinant mouse CXCL16 or CXCL16-Fc. (Fig. S5). Taken together, these data imply that the CXCL16-CXCR6 system appears to be dispensable for the migration and function of colitogenic T cells.

CXCR6⁻CD4⁺ T cells but not CXCR6⁺CD4⁺ T cells can Transfer Colitis

To further assess the roles of CXCR6⁺ and CXCR6⁻ T cells in the pathogenesis of colitis, we next performed adoptive retransfer of LP CXCR6⁺ and CXCR6⁻ CD4⁺ T cells recovered from the inflamed colon. To avoid carry over of regulatory T cells into the recipient mice, we eliminated the CD25^high population from the donor cells (Fig. 6A). Unexpectedly, Rag1^−/− recipients of the CXCR6⁻ subset exhibited progressive body weight loss with clinical symptoms of colitis to a similar extent as the recipients of splenic CD45RB^high naïve CD4⁺ T cells. On the other hand, the recipients of the CXCR6⁺ subset exhibited mild body weight loss at 1 week post-transfer; however, they recovered quickly and remained healthy for the 8-week duration of the study (Fig. 6B). The colon/body weight ratio of the recipients of the CXCR6⁻ subset was significantly higher than the CXCR6⁺ subset at 8 weeks (data not shown). Histopathological analysis of the distal colon tissues from the two groups indicated that the transfer of the CXCR6⁻ but not CXCR6⁺ subset induced chronic inflammation similar to that observed in CD45RB^high-transferred Rag1^−/− recipients (Fig. 6C). These observations indicate that only the CXCR6⁻ subset retains the ability to transfer the disease. Rag1^−/− recipients reconstituted with CXCR6⁺ cells contained a limited number of CD4⁺ T cells in SP, MLN, and colon LP that retained CXCR6 expression (Fig. 6D, 6E). In sharp contrast, the CXCR6⁻ cells retransferred into Rag1^−/− recipients vigorously expanded and the majority of them upregulated CXCR6 expression. The conversion of retransferred CXCR6⁻ into CXCR6⁺ cells was most prominent in LP among the tissues tested (Fig. 6E). The CD4⁺ T cells that acquired CXCR6 expression predominantly produced effector cytokines such as IFN-γ and IL-17A (Fig. 6F), suggesting that CXCR6⁺ cells generated in situ from CXCR6⁻ cells mediate the colonic inflammation.

The fact that recipients of retransferred CXCR6⁺ cells failed to develop the disease raised the possibility that this could be a population of terminally differentiated effector cells with little proliferative activity. Indeed, the CXCR6⁺ subset was less proliferative compared to the CXCR6⁻ counterpart upon in vitro restimulation (Fig. 6G). Collectively, these observations indicate that the CXCR6⁻ cells actively expand, giving rise to CXC6⁺ effector T cells that are responsible for the persistence of the chronic colitis.